Pollo AI turns creative ideas into campaigns with OpenAI

Pollo AI ⁠ (opens in a new window) was founded on the premise that people should be able to turn their creative ideas into finished images and videos without needing production expertise. Even with generative AI, creators and marketers can struggle to navigate different tools and choose between models to get the result they want. Some may even abandon their projects in frustration.

“We want to do for video what Canva did for design. We want to make it easy enough that more people can actually use it, not just the ones who already know how,” explains Bill Zhu, the founder and CEO of Pollo AI. Today, Pollo AI serves more than 26 million users who use the platform to create videos, ads, and visual content.

The team’s first breakthrough was video templates. Now, 30% of creator sessions start from these templates or trending-format workflows. Building on this success, the company’s latest offering is Pollo Agent. It uses OpenAI models, including GPT‑6 Astra and GPT‑Image‑2.5, to provide a fully guided workflow for video creation, including this post-apocalyptic adventure from a Pollo AI user.

“We think the AI shift in content creation will favor products that make creation more repeatable from the start, with stronger formats, less trial and error, and a simpler workflow around the models themselves,” he says.

Pollo Agent interprets the user’s rough ideas and visual references, and then develops storylines, scene flows, and scripts. CTO Peter Zhou says OpenAI models stood out for their ability to understand ambiguous creative inputs and their multimodal capability, sharing: “GPT‑5.6 performed especially well there.” The team also said they value the OpenAI API’s stability and release cadence.

To balance reasoning depth, speed, and cost, Pollo AI uses GPT‑5.6 for routing and GPT‑6 Astra for more challenging work, such as developing a narrative or planning scene revisions. This model mix lets the system match the level of reasoning to the task based on the user’s prompt. Pollo AI reports that users spend more than 50% less time choosing or switching models, giving them more time for creativity.

Pollo AI tests generative models against real user prompts and workflows, evaluating output quality, consistency, stylistic range, speed, and cost. GPT‑Image‑2.5 was selected to handle image generation, helping creators make posters, storyboards, and campaign visuals.